SIMULASI DAN ANALISIS ERROR KOMPUTASI FFT WINOGRAD 16-TITIK MENGGUNAKAN XILINX ISE 10.1I

Abstract

Weakness data processing using analog processors are less efficient because if there is an error in the design of a system
using an analog processor, the hardware of the system should be redesigned. Processing of analog signals using digital processor has several advantages such as efficient and easy to modify the system are made, without requiring hardware redesign as well as in the design of analog systems. Errors in the design of digital circuits in the processor only requires modification program, without having to change the hardware. Modifications can be done anywhere, without demanding we must be in the laboratory. Fast Fourier Transform (FFT) is a computational algorithm that is used for digital data processing such as in the field of image, music, and satellite. Winograd FFT 16 point multiplier, multiplier consists of 6 pieces in the form of fractions. Design of FFT multiplier in Xilinx ISE 10.1i could do with changing fractions menajdi integers and then converted into a number by logic 1 and 0. Changing fractions into an integer will give different computational results compared with the original computation using a multiplier of FFT. The shift results enormous computing will result in lost / loss information can be conveyed from the processed data. To that end, researchers tried to simulate and analyze the computational error rate of 16 point Winograd FFT processor using Xilinx ISE 10.1i. The results showed average percentage error computational simulations using FFT 16 point Xilinx Ise 10.1i compare with Matlab is 6.67% (first trial) and 4:48% (second trial). This error occurs because the processor does not allow the FPGA digital display numbers in a real number.